Optimizing Cluster Defect Identification in 300 mm Silicon Wafers: Impact of Spot Size and Size Threshold in Surface Scanning Inspection Systems

This study investigates the effects of laser spot size and sizing threshold parameters in Surface Scanning Inspection Systems (SSIS) on defect cluster identification in 300 mm silicon wafers. Reducing the laser spot size enhances the detection of smaller point defects, increasing cluster counts by up to 45% depending on the scan mode. Similarly, lowering the sizing threshold improves point defect capture rates, raising cluster probability by over 30%. However, these optimizations involve trade-offs: smaller spot sizes limit the search distance, while lower thresholds encourage cluster merging, resulting in a net decrease in identified clusters by 10%-20% in some cases. These findings underscore the need for a precise balance between sensitivity and resolution in SSIS configurations to achieve effective defect management. This work offers critical insights for optimizing silicon wafer quality, an essential factor in advanced integrated circuit manufacturing. Furthermore, it establishes a foundation for integrating machine learning-based defect feature analysis, paving the way for next-generation automated and predictive inspection systems in semiconductor fabrication. By addressing the challenges of defect clustering and providing data-driven solutions, this study contributes to the advancement of more reliable and efficient inspection technologies for the evolving demands of the semiconductor industry.